Our Projects
Overview
The projects are connected by a local-first approach to AI, automation, and developer infrastructure. Some are public tools, some are applied AI work, and some support sensitive legal and family-law workflows.
Capability Areas
- Security / Runtime: Secrets Kit keeps API keys, passwords, and tokens out of
.envfiles, shell history,argv, andlaunchdplists. - Operations / Infrastructure: LLM Ops Kit supports running, deploying, debugging, and maintaining self-hosted AI stacks.
- Developer Tooling: UnicodeFix cleans AI-generated text for code, docs, posts, and workflows that need plain ASCII.
- Applied AI: Case Analytics applies AI and data analysis to sensitive legal and family-law documents.
- Model Engineering: Current Apple Silicon model-conversion and inference-performance work is moving into MLX Forge, which remains private while it clears its initial milestones.
Together, these form a practical ecosystem: secure runtime handling, local operations, developer utilities, and applied analysis for sensitive data.
Internal Projects
Internal projects focus on applied AI for sensitive documents, including legal case materials, interpersonal communications, and family-law analysis. Parental alienation detection remains an important specialized vertical.
|
Case AnalyticsAI-powered analysis tool for legal case documents, focusing on parental alienation detection and pattern recognition in high-conflict divorce cases. |
|
Parental Alienation AwarenessParental Alienation Awareness: A data-driven look at the scope, impact, and damage to children, families, and society. |
Open Source Projects
Open-source projects support the broader local-first stack: secrets hygiene, AI operations, text cleanup, model runtimes, environment management, and developer automation.
|
LogGPT and LogGPT PlusSafari extension for exporting ChatGPT conversations as JSON, with an optional Plus upgrade that preserves generated and uploaded artifacts in a ZIP archive. |
|
UnicodeFixEvidence-based local audit and safe cleanup for hidden Unicode, provenance, and configured watermark signals. |
|
Secrets KitSimple, secure CLI for secrets and PII. Store values in Keychain, keep metadata in a local registry, and export runtime env values without putting secrets in git. |
|
LLM Ops KitOperational toolkit for running, deploying, debugging, and maintaining self-hosted AI stacks across hosts. |
|
Extract ChatExtract ChatGPT sessions from downloaded JSON into HTML and Markdown with metadata, also chunking for ML pipeline. |
|
WebReader for SafariA local-first Safari extension that reads selected text or complete articles aloud through your own text-to-speech system. |
|
ConnectomeAIConnectomeAI: experiments in mapping and orchestrating AI tools, workflows, and knowledge graphs. |
|
VenvUtilPython virtual environment management functions and script to build and manage performance, compatibility, and regression test venv builds mostly for AI |
|
TorchDeviceModule contains class TorchDevice abstracting CUDA/MPS/CPU, intercepting, and redirecting calls to the available device to assist in porting code from CUDA to MPS. |